Abstract:A frequent kind of arthritis, knee osteoarthritis is characterized by sclerosis, joint space narrowing, osteophyte growth, and bone deformities that can be seen on radiographs. Radiography is the most affordable and widely accessible method, and it is considered to be the best. The Kellgren and Lawrence (KL) grading technique is used to classify X-ray pictures in accordance with the progression of osteoarthritis from normal to severe. Degeneration of osteoarthritis in the knee can be slowed down by early identification, which can aid in early treatment. Regretfully, in an effort to enhance the performance of their models, the majority of currently used methods either combine or eliminate confusing grades. The objective of this research is to present an approach by leveraging an ensemble of CNN models, specifically MobileNet, ResNet, and AlexNet architectures. The choice of using a Convolutional Neural Network (CNN) for knee osteoarthritis classification is driven by its capacity to leverage deep learning techniques for medical image analysis. CNNs excel at feature extraction from medical images, making them ideal for identifying subtle patterns indicative of osteoarthritis. This approach improves the potential to automate diagnosis, reduce human error, and patient outcomes by enabling timely intervention, underscoring its relevance in the realm of medical image analysis. An Osteoarthritis Initiative (OAI) based dataset of knee joint X-ray images is chosen for this study. The dataset was split into the training, testing, and validation set with a 7.5: 1.5: 1 ratio. Our results shows that the ensemble approach significantly outperforms individual model predictions, achieving an accuracy of 96%. This improvement underscores the potential of using deep learning ensembles in medical image analysis, offering enhanced diagnostic processes in KOA classification.
Keywords:Knee Osteoarthritis (KOA), Osteoarthritis dataset, CNN, AlexNet, ResNet, Mobile Net, Ensemble model
|
DOI:
10.17148/IJARCCE.2024.134155
